Evaluating the potential of ChatGPT-4 in ophthalmology: The good, the bad and the ugly.

There is growing interest nowadays for artificial intelligence (AI) in all medical fields. Beyond the direct medical application of AI to medical data, generative AI such as "pre-trained transformer" (GPT) could significantly change the ophthalmology landscape, opening up new avenues for enhancing precision, productivity, and patient outcomes. At present, ChatGPT-4 has been investigated in various ways in ophthalmology for research, medical education, and support for clinical decisions purposes. This article intends to demonstrate the application of ChatGPT-4 within the field of ophthalmology by employing a 'mise en abime' approach. While we explore its potential to enhance the future of ophthalmology care, we will also carefully outline its current limitations and potential risks.

Low-power versus medium-power toric intraocular lenses in cataract surgery, about 110 eyes.

PURPOSE: To analyze the refractive predictability of low-power and medium-power toric intraocular lenses (IOL) during cataract surgery. METHODS: A retrospective case series. Patients with low and moderate degrees of astigmatism were treated with either an Alcon AcrySof IQ toric monofocal lens (SN6AT2 or SN6AT3) or a Zeiss toric monofocal lens (709 MP + 1 D or 709 MP + 1.5 D). They were divided into two groups: group 1 with low-power toric IOL (n = 40, SN6AT2 and 709 MP + 1), group 2 with medium-power toric IOL (n = 70, SN6AT3 and 709 MP + 1.5 D). Eyes were evaluated before and 1 month after surgery. Uncorrected distance visual acuity (UDVA), corneal and total astigmatism were compared pre- and postoperatively. Vector astigmatism analysis was evaluated using the Alpins method. IOL alignment was checked postoperatively at slit lamp under pupil dilatation and on photos using Eyesuite® software (Luneau®). RESULTS: More than three-quarters of the patients in each group attained a visual acuity of at least 0.1 LogMAR (0.8) without spectacles. The reduction of total astigmatism was significant in the two groups. The residual refractive astigmatism did not differ between the two groups (P = 0.64) and was less than 0.4 D. More than 80% of patients (82.5 vs 84.3%) presented a postoperative refractive spherical equivalent at ±0.50 D (P = 0.8). Vectorial astigmatism was significantly different in the two groups (0.43 D in group 1, vs 0.27 D in group 2; P = 0.03). CONCLUSION: This is the first study comparing the low- versus medium-power toric IOLs, the most widely used. This study suggests very good refractive results both with low- and medium-power toric IOLs with a single surgical procedure.